Mandibular gland pheromone contents in workers

Abdullahi A. YUSUF, Christian W. W. PIRK, Robin M. CREWE

Social Insects Research Group, Department of Zoology and Entomology, University of Pretoria, Private Bag X20,Hatfield 0028, Republic of South Africa

Received 9 May 2014 – Revised 29 November 2014 – Accepted 12 December 2014

Abstract – Secretions from the mandibular glands of honeybees have been studied extensively, with those ofqueens dominated by ώ-9 fatty acids and ώ-10 fatty acids dominating those of non-laying workers. Apis melliferaadansonii (Latreille) is one of the widely distributed subspecies of African honeybees. However, its mandibulargland pheromones have not been analysed previously. Using gas chromatography, we analysed the composition ofmandibular gland pheromones in workers and queens ofA. mellifera adansonii fromNigeria. Qualitatively, workersand queens have similar pheromone profiles to those previously reported in other African subspecies of honeybees.We found 9-ODA and high amounts of its precursor 9-hydroxy-2 (E )-decenoic acid (9-HDA) in workers, thusshowing that they produce queen-like signals under queen-right conditions. We also found geographic variation inthe pheromone profiles and morphometric characters of these workers, suggesting different pheromone andmorphoclusters from the different ecological and climatological regions inhabited by A. m. adansonii in Nigeria.

African honeybees / queen-right / mandibular gland secretions / pheromone clusters / morphoclusters

1. INTRODUCTION

Social organisation driven by reproductivedominance is a key characteristic of eusocialityin insects. In honeybees, this is mainly controlledthrough effective communication between thequeen and workers in the colony. The honeybeecolony provides a locale in which cues, rangingfrom the temperature produced by active workersin the brood nest (Basile et al., 2008) to broodpheromones emitted by the open brood (Le Conteet al., 1990) and glandular secretions (especiallythose from the mandibular glands relating toreproductive dominance Crewe and Velthuis,1980), can interact and determine responses bythe members of the colony.

Queen pheromone components function pri-marily to keep the colony in a queen-right condi-tion by signalling the presence of a viable queen toother workers throughout the colony or over shortdistances. Chemical signals from the mandibulargland secretions of honeybee queens (Apismellifera ) have been implicated in the control ofbehavioural and physiological activities such aseliciting retinue behaviour among workers(Slessor et al., 1988), inhibiting emergency queenrearing as well as ovary activation in workers(Butler, 1959). The constituents of this gland havebeen studied in detail, and the active componentsidentified as methyl p -hydroxybenzoate, 9-oxo-2(E )-decenoic acid (9-ODA) (which constitutesabout 80 % of the secretions in mated queens),4-hydroxy-3-methoxyphenylethanol (HVA),(R ,E )-9-hydroxy-2-decenoic acid (9-HDA),(S,E )-9-hydroxy- 2-decenoic acid (9-HDA), 10-hydroxy-decanoic acid (10-HDAA) and 10-hydroxy-2 (E )-decenoic acid (10-HDA) (Slessoret al., 1988;Winston et al., 1989). 9-ODA, the two

Corresponding author: A. Yusuf,aayusuf@zoology.up.ac.zaManuscript editor: Peter Rosenkranz

Apidologie (2015) 46:559–572 Original article* INRA, DIB and Springer-Verlag France, 2015DOI: 10.1007/s13592-014-0346-6

and queens of Apis mellifera adansonii

enantiomers (R and S -9HDA) and the two aro-matic compounds (HOB and HVA) are referredcollectively as the queen mandibular pheromones(QMPs) reviewed in Pirk et al. (2011). These fivecompounds act together to maintain reproductivedominance and elicit worker retinue behaviour.Keeling et al. (2003) identified four other compo-nents that are not of mandibular gland origin,namely methyl (Z )-octadec-9-2-en-1-o l (methyloleate), (E )-3-(4-hydroxy-3-methoxyphenyl)-prop-2-en-1-o l (coniferyl alcohol), hexadecane-1-o l and (Z9 , Z12 , Z15 )-octadeca-9, 12, 15-trienoic acid (linolenic acid), which act in synergywith QMPs to elicit retinue behaviour. This leadsto a terminology shift from QMP to queen retinuepheromones (QRP) (Slessor et al., 2005). In addi-tion, the mandibular glands of non-laying workersproduce large amounts of (E )-10-hydroxy-2-decenoic acid (10-HDA) known as worker sub-stance, which is also found in royal and workerjellies and has been implicated in queen determi-nation in larvae (Spannhoff et al., 2011).

Africa is home to 10 (Hepburn and Radloff,1998) or 11 (Ruttner, 1988) morphoclusters(subspecies) in the Apis mellifera complex thathave distinct geographic distributions andmorphol-ogies. However, we only know the compositionand spectrum of the mandibular gland fatty acidsfrom four subspecies, i.e. capensis , scutellata ,intermissa and saharensis (Crewe, 1982; Creweand Moritz, 1989; Hepburn and Radloff, 1996).This knowledge has shown that the compositionof pheromone bouquets from queens and workersis subspecies specific, especially in relation to theability of Apis mellifera capensis workers to be-come facultative social parasites and mimic thepheromone bouquet of queens in their host colonies(Crewe and Velthuis, 1980). Such variation in thecomposition of an important primer and releaserpheromone which affects both worker and queenbehaviour in the hive needs to be understood notonly in races of African honeybees but also in otherraces. An understanding of these differences isimportant for the management and conservationof different races of honeybees on which sustain-able apicultural practices will be dependent.

The West African honeybee A. melliferaadansonii (Latreille) is found along the WestAfrican coast of Senegal, Chad, Nigeria, far south

Cameroon (Hepburn and Radloff, 1998), the Con-go basin as well as Angola to the South (Ruttner,1988). This makes it one of the two subspecieswith the widest geographic range among Africansubspecies. However, despite its wide distributionand being the first African subspecies describedby Latreille (in Ruttner (1992)), there is no infor-mation on the chemistry and composition of man-dibular gland pheromones in A. melliferaadansonii from any of its native range.

To gain an insight into its chemical signals andhow these vary along an ecological gradient, wecompared the mandibular gland pheromone com-position of A. mellifera adansonii queens andworkers from different regions of Nigeria. We sam-pled workers from three regions and queens fromtwo regions of Nigeria (North West (Sudan savan-nah), North Central (Guinea forest savannah) andSouth West (rainforest)) to examine the variation inmandibular gland pheromone composition as wellas bee morphology across these ecological zones.We hypothesise that mandibular gland pheromonesin both workers and queens of A. melliferaadansonii will be the same to those of the A branchof the mellifera complex (Ruttner 1988) especiallyApis mellifera scutellata . In addition, we expectthat there will not be any geographic variation inthe proportions and amounts of A. melliferaadansonii mandibular gland pheromones and mor-phometric characters across Nigeria.

2. MATERIALS AND METHODS

2.1. Honeybees

WorkersWorkers (total=383) of A. mellifera adansoniiwere sampled from frames in 30 queen right coloniesfrom apiaries at Jaja (11° 15′ N, 07° 39′ E), TasharFulani (11° 17′ N, 07° 40′ E), Shika Dam (11° 08′ N,07° 40′ E), Unguwan Dan Asabe (11° 09′N, 08° 20′ E),Zabi (11° 18′ N, 07° 43′ E) in Zaria, Kaduna State(North West), a commercial apiary in Roguwa (08°49′ N, 07° 43′ E) Keffi, Nasarawa State (North Central)and two apiaries from Iwo (07° 33′ N, 04° 11′ E) andOsun and Ijebu-Ode (06° 46′ N, 03° 56′ E) Ogun States(South West), Nigeria (Figure 1) between December2011 and January 2013. The colonies sampled werefrom sites characterised by stationary smallholder oper-ations. The workers were randomly sampled, placed in

560 A.A. Yusuf et al.

clean labelled, perforated plastic vials and immobilisedon ice. The heads were then decapitated and placed intoa clean 2 mL vial containing dichloromethane(ChromSolv® grade for HPLC Sigma-Aldrich, St. Lou-is, MO, USA) and refrigerated until analysis.

Queens Naturally mated, egg laying queens were sam-pled from 17 colonies (13 from North West and 4 fromNorth Central) in apiaries where queen rearing and con-trolledmatingwere not practised.Unfortunately, no queenswere collected from apiaries in the South West since thebeekeepers were reluctant to sacrifice their queens. Queenswere collected from colonies, and the heads sampled in asimilar way as described for workers above.

2.2. Extraction of mandibular glandpheromones from heads

Heads were extracted in 200 μL of dichloromethane(HPLC grade, Sigma-Aldrich) following the methods

described in the studies of Dietemann et al. (2006) andZheng et al. (2010). Prior to gas chromatography, eachextract was divided into two and half was stored as abackup for further analysis or confirmation (if need be).The other half (∼100 μL) was transferred into a vial andevaporated to dryness under a gentle stream of charcoal-filtered nitrogen gas. The residue was then re-dissolvedin 10 μL of internal standard solution (containing 1 mgof octanoic acid and 1 mg of tetradecane in 4 ml dichlo-romethane) to which 10 μL of bis-(trimethylsily)trifluoroacetamide (BSTFA) was added to derivatisethe fatty acids.

2.3. Gas chromatography

One microlitre of the derivatised extract from abovewas injected in splitless mode into an Agilent 6890 gaschromatograph fitted with a 25 m×0.20 mm×0.33 μmHP1-methyl silicone-coated fused silica capillary columnand an FID detector. The carrier gas was helium at a flowrate of 1 mL/min and oven temperature-programmed as

Figure 1.Map of Nigeria showing apiaries where samples were collected from Jaja, Tashar Fulani, Unguwar DanAsabe in Zaria Kaduna state North West (circles ); Roguwa, Keffi Nasarawa state North Central (triangle ); and IwoOsun state and Ijebu-Ode, Ogun state South West (squares ) Nigeria.

Mandibular gland pheromones of Apis mellifera adansonii 561

follows: 50 °C at 50 °C/min to 100 °C, then increased at3 °C /min to 220 °C and then held at 220 °C for 10 min(modified after Simon et al., 2001; Dietemann et al.,2006; Tan et al., 2012). Chromatograms were recordedand the peak areas quantified using Chemstation® soft-ware. Six of the major components from mandibularglands of honeybees that had been shown to elicit bothbehavioural and physiological responses (namely methylp -hydroxybenzoate, 9-oxo-2 (E )-decenoic acid (9-ODA), 4-hydroxy-3-methoxyphenylethanol (HVA), 9-hydroxy-2 (E )-decenoic acid (9-HDA), 10-hydroxy-decanoic acid (10-HDAA) and 10-hydroxy-2 (E )-decenoic acid (10-HDA)) were identified based on com-parison with retention times of synthetic standards.Quantification was achieved by comparing the relativemass ratios (RMR) of each of these compounds in astandard solution mixture (containing ∼1 mg of each in4 mL dichloromethane) relative to the RMR oftetradecane. We did not separate the enantiomers of 9-HDA in our analysis and, therefore, reported both enan-tiomers together throughout the study. As a check forshifts in retention times of compounds between GC runs,standard mixtures were run before each batch of headextracts and then after every 25 to 30 samples. Allstandards and reagents with the exception of 9-ODAand 9-HDA (PheroTech Inc.) were obtained from Sig-ma-Aldrich.

2.4. Morphometrics

Ten worker bees (not those used for pheromoneanalysis) were sampled from each of the 30 coloniesused in this study. The bees were preserved in 90 %ethanol and later dissected according to the methods ofRuttner (1988). The following seven morphometriccharacteristics based on the Oberursel standard list(Ruttner, 1988) were measured: forewing length (FL),forewing width (FW), hindwing length (HL), hindwingwidth (HW), proboscis length (PL), number of land-marks on the forewing (NF) and number of landmarkson the hindwing (NH) were measured. All together, 300worker bees were analysed and the values expressed inmillimeters, except for NF and NH which were countdata.

2.5. Statistical analyses

Since data for both percentage composition andamoun t s were no t no rma l ly d i s t r ibu ted ,

non-parametric statistical tests were applied.Kruskal-Wallis ANOVA (KWA) and median testwith multiple comparisons were applied to deter-mine differences in the individual as well as totalMDG components in workers between regions,with regions as the independent (grouping) vari-able and the MDG components as dependent var-iables. Since no queens were sampled from SouthWest, a Mann-Whitney U test (MWU) was usedto test for significance in MDG components be-tween queens from North Central and North West.Cluster analyses (Johnson and Wichern, 1998)were performed using the means for both propor-tions and amounts of pheromones in order todetermine the linkage/Euclidean distances betweenworkers from the three regions. Multivariate sta-tistical analyses, including principal componentanalysis, canonical analysis and linear discriminantanalysis, were employed on the morphologicalcharacters to classify colonies in relation to theirgeographic region. The population means betweengroups were compared using Wilk’s lambda testand its distribution approximated by the F distri-bution (Johnson and Wichern, 1998). Results un-less otherwise stated are presented as means±thestandard errors of the mean. All statistical testswere carried out using the software STATISTICA12 (StatsSoft Inc. USA), and α was set at 0.05.

3. RESULTS

3.1. Proportion of MDG componentsin workers and queens

The proportions of the major MDG com-ponents from the head extracts of workersand queens of A. mellifera adansonii arepresented in Figure 2(a (i), a (ii)) respective-ly. All the major components with the excep-tion of HVA were present in workers, with10-HDA and its precursor 10-HDAA domi-nating the profile at 43.49±1.22 and 23.99±0.87 %, respectively, followed by HOB, 9-HDA and 9-ODA respectively. The averageproportions of 9-ODA and 9-HDA that arethe major components of queen signals were17.26±0.83 % [Figure 2(b (i))].

In the case of the queens, all six major andbehaviourally active MDG components were

562 A.A. Yusuf et al.

present. The 9-ODA was the most abundant at63.64±2.74 %, followed by its precursor 9-HDA, HOB, then the two worker components10-HDAA and 10-HDA, with HVA as a minorcomponent [Figure 2(a (ii))].

3.2. Absolute amounts ofMDG componentsin workers and queens

Absolute amounts were very variable int h e mand i b u l a r g l a n d s o f wo r k e r s[Figure 2(b (i))] with total absolute amountsaveraging 7.0±0.55 μg per bee. Even thoughthe worker component 10-HDA was highlyvariable at 4.53±0.49 μg, it was the mostabundant component in the mandibularglands of workers. This was followed by itsprecursor 10-HDAA (1.03±0.06 μg), HOB(0.71±0.04 μg) and some queen substance9-ODA (0.09±0.01 μg) and its precursor 9-HDA (0.66±0.05 μg).

In queens, the total absolute amount was229.37±25.54 μg per queen with 9-ODA at the

level of 146.44±17.74 μg, its precursor 9-HDA(39.17±5.09 μg) with HVA as the minor compo-nent at 1.54±0.28 μg [Figure 2(b (ii))].

3.3. Regional variability in the pheromoneprofiles of workers

Results from the three regions (North Central,North West and South West Nigeria, Table I)showed that the proportions of 9-ODA, 9-HDA,10-HDAA and 10-HDAwere different (KWA: H(2, N=383), P<0.05) while that of HOB were notsignificantly different (KWA: H (2, N =383),P=0.3495) irrespective of the region sampled.Profiles of workers from SouthWest had the most(54.42±1.62 %) 10-HDA (workers substance) incomparison to those from North Central andNorth West respectively (Table I). Interestingly,those workers from the North West and NorthCentral had proportionally more of the queensubstance 9-ODA and its precursor 9-HDA com-pared to those from the South West (Table I).Based on the mean proportions of their

Figure 2. Percentage composition (a ) and amount (μg) (b ) of mandibular gland pheromones in workers (blackbars ) and queens (open bars ) of Apis mellifera adansonii . Please note the differences in scales for workers andqueens of the amounts in micrograms. Mandibular gland components are: HOB=p -hydroxybenzoate, 9-ODA=9-oxo-2 (E )-decenoic acid, HVA=4-hydroxy-3-methoxyphenylethanol, 9-HDA=9-hydroxy- 2 (E )-decenoic acid,10-HDAA=10-hydroxy-decanoic acid and 10-HDA=10-hydroxy-2 (E )-decenoic acid. Error bars are standarderrors of the means, n=number of samples.

Mandibular gland pheromones of Apis mellifera adansonii 563

pheromones, workers from North Central andNorth West formed a distinct pheromone clusterfrom those in the South West (Figure 3).

The variation in absolute pheromone amountsamong A. mellifera adansonii workers from thethree regions is shown in Figure 4. This wassignificantly different between the regions for allthe worker components as well as the total abso-lute amount (KWA: H (2, N=383), P<0.05). The

pattern of variation shown is in the following orderNorth Central>South West>North West for theamount of 9-ODA, 9-HDA and 10-HDAA, whilethe amount of HOB, 10-HDA and total amountsdecreased in amount in the following order: SouthWest>North Central>North West respectively(Figure 4), thereby exhibiting an upward trend,increasing altitudinally as one travels through thehinterland (i.e. inwards from the South to the

Table I. Relative proportions (mean±SE) of mandibular gland pheromones from head extracts of Apis melliferaadansonii workers and their differences between the North Central (NC), North West (NW) and South West (SW)regions of Nigeria.

Compounds North Central (NC)Mean proportion±SE

North West (NW)Mean proportion±SE

South West (SW)Mean proportion±SE

Kruskal-WallisANOVA H, p

Differences

HOB 13.24±1.49 15.9±1.10 15.04±0.91 H=2.103,p= 0.3495

None

9-ODA 3.83±0.72 3.30±0.40 1.27±0.15 H=43.470,p= 0.0000

NC, NW,SW

9-HDA 15.65±1.52 16.8±0.90 11.52±0.74 H=18.607,p= 0.0001

NC, NW,SW

10-HDAA 23.95±1.84 28.7±1.40 17.75±0.93 H=24.460,p= 0.0000

NC, NW,SW

10-HDA 43.33±3.26 35.3±1.70 54.42±1.62 H=58.950,p= 0.0000

NC, NW,SW

HOB=p-hydroxybenzoate, 9-ODA=9-oxo-2 (E )-decenoic acid, HVA=4-hydroxy-3-methoxyphenylethanol, 9-HDA=9-hydroxy-2 (E )-decenoic acid, 10-HDAA=10-hydroxy-decanoic acid and 10-HDA=10-hydroxy-2 (E )-decenoic acid

8 10 12 14 16 18 20 22

Linkage/Euclidean Distance

South west

North west

North central

Figure 3. Dendrogram from cluster analysis of the workers from North Central, NorthWest and SouthWest Nigeriabased on mean proportions of each chemical from the head extracts showing the two pheromone clusters (one forworkers from North Central and North West, and the other for workers from South West). The Euclidean distancesare 9.8, 13.7 and 22.8 for North Central, North West and South West, respectively.

564 A.A. Yusuf et al.

North; see Figure 1). Mean total absolute amountswere 9.33±1.15 μg per bee for South West, 6.93

±1.18 μg for North Central and 5.31±0.73 μg perbee from North West respectively. A cluster

HOB

NC SW NW

Region

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Am

ou

nt

(µg)

9-ODA

NC SW NW

Region

-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Am

ou

nt

(µg

)

9-HDA

NC SW NW

Region

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.510-HDAA

NC SW NW

Region

-1

0

1

2

3

4

5

6

7A

mo

un

t (µ

g)

10-HDA

NC SW NW

Region

-2

0

2

4

6

8

10

12

14

16

18

20

22

Am

ou

nt

(µg

)

Total

NC SW NW

Region

-5

0

5

10

15

20

25

30

35

Am

ou

nt

(µg

)

Figure 4. Amount of five major components of the mandibular gland pheromones (HOB, 9-ODA, 9-HDA, 10-HDAAand 10-HDA) and total absolute amounts in workers of A. mellifera adansonii from three geographical regions ofNigeria. NC North Central (n=45), SW South West (n=145), NW North West (n=193). = median, = 1st and 3rdquartiles, = non-outlier range (minimum and maximum), = outliers.

Mandibular gland pheromones of Apis mellifera adansonii 565

analysis on the absolute amounts grouped theworkers into two distinct pheromone clusters sim-ilar to the one for proportions (Figure 5).

3.4. Regional variability in the pheromoneprofiles of queens

Proportional compositions of the MDGs inA. mellifera adansonii queens were not signif-icantly different (MWU test P >0.05) betweenthose queens from the North Central and NorthWest (Figure 6). The queen substance 9-ODA,which dominated the pheromone profiles, was70.12±9.42 and 61.6±2.2 % in queens fromNorth Central and North West, respectively(Figure 6a).

Absolute amounts of the MDG compo-nents in queens from the two regions areshown in Figure 6b. No marked differencesin the absolute amounts were observed be-tween the queens from different regions.Overall comparisons between the queensshowed that 9-HDA (the precursor of thequeen substance 9-ODA) was slightly differ-ent (MWU test, Z =−1.9814, P =0.047).However, comparatively queens from North

West possess more MDG pheromones(Figure 6b) with a mean total of 242.44±33.96 μg per queen to 187±43.21 μg perqueen from the North Central.

3.5. Morphometric analysis

The measured morphometric characters areshown in Table II. Length of forewing (FL)and the number of landmarks on both foreand hind wings were not different betweenworkers from all regions (KWA, P =0.119,0.2117 and 0.06 respectively). Width of fore-wings (FW), length of hind wings (HL) andproboscis (PL) were different across all re-gions (KWA, P =0.0001, 0.0001 and 0.0002respectively). By contrast, width of hindwings (HW) only differed in workers fromthe North West (KWA, P =0.0007).

The principal component analysis of themorphometric characters for the worker hon-eybees isolated these five factors: FL, FW,HL, HW and PL, with eigenvalues >1 whichaccounted for 79.59 % of the variance in thedata. A stepwise discriminant analysis usingthe colony means of the morphometric

1.5 2.0 2.5 3.0 3.5 4.0

Linkage/Euclidean Distance

South west

North west

North central

Figure 5. Dendrogram from cluster analysis of the workers from North Central, NorthWest and SouthWest Nigeriabased on mean absolute amounts (μg) of each chemical compound and the total amounts from the head extractsshowing the two pheromone clusters (one for workers fromNorth Central and NorthWest, and the other for workersfrom South West). The Euclidean distances are 1.85, 3.81 and 5.29 for North Central, North West and South West,respectively.

566 A.A. Yusuf et al.

characters separated the workers into threeclusters (Figure 7). From South West,87.5 % of colonies were classified correctly,with posterior probabilities of P =1.0 for sev-en colonies and 0.08 for one colony (whichwas incorrectly classified). All colonies fromthe North Central and North West (100 %)were classified correctly with posterior prob-abilities of P =1.0. Mahalanobis distance D 2

between the clusters was 132.29 for SouthWest, 66.53 for North Central and 37.93 forNorth West. Standard coefficient of canonicalvariance was 95.10, 14.29 (Wilk’s lamb-da=0.0126, F (10, 46)=36.611, P <0.0001).

4. DISCUSSION

One of the main findings here is that the man-dibular gland pheromone profiles in workers andqueens of A. mellifera adansonii , a subspecies ofAfrican honeybees with a wide distribution rangeon the African continent, have a high degree ofvariability. We found in these head extracts all themain and behaviourally active compounds report-ed earlier from the mandibular glands of differentcastes in honeybees (Butler, 1959; Slessor et al.,1988; Plettner et al., 1997; Slessor et al., 2005).

The profiles of A. mellifera adansonii workerswere dominated by 10-HDA and its precursor 10-

Figure 6. Percentage composition (a ) and mean amount (μg) (b ) of the individual mandibular gland pheromones fromheads of queens of Apis mellifera adansonii from North Central (open bars ) and North Western (closed black bars )Nigeria. Mandibular gland components are the following: HOB=p -hydroxybenzoate, 9-ODA=9-oxo-2 (E )-decenoicacid, HVA=4-hydroxy-3-methoxyphenylethanol, 9-HDA=9-hydroxy- 2 (E )-decenoic acid, 10-HDAA=10-hydroxy-decanoic acid and 10-HDA=10-hydroxy-2 (E )-decenoic acid. Error bars represent standard errors of the means.

Mandibular gland pheromones of Apis mellifera adansonii 567

Table II. Means and standard errors of measured morphometric characters (lengths are in mm) of A. mellifera.adansonii workers from North Central (NC), North West (NW) and South West (SW) Nigeria.

Charactera North Central(NC) Mean±SE

North West(NW) Mean±SE

South West(SW) Mean±SE

Kruskal-WallisANOVA H, P

Differences

N=40 N=180 N=80

FL 6.83±0.06 6.90±0.03 6.98±0.04 H=4.38,P=0.1119

None

FW 2.17±0.03 2.45±0.05 2.27±0.04 H=21.11,P=0.0000

NC, NW, SW

HL 4.68±0.04 4.98±0.01 4.83±0.03 H=18.37,P=0.0001

NC, NW, SW

HW 1.32±0.02 1.37±0.03 1.31±0.02 H=14.62,P=0.0007

NW

PL 2.62±0.02 1.79±0.05 1.46±0.12 H=16.88,P=0.0002

NC, NW, SW

NF 13.00±0.00 13.00±0.00 13.00±0.00 H=3.05,P=0.2117

None

NH 6.00±0.00 6.00±0.00 6.00±0.00 H=5.68,P=0.06

None

aMorphometric characters were explained in the text

N number of workers

-3 -2 -1 0 1 2 3 4 5

Discriminant function 1

-2.0

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

Dis

crim

inan

t fun

ctio

n 2

North West

South West

North Central

Figure 7. Discriminant analysis plot showing discriminant functions 1 and 2 using the colony means of themorphometric data. Triangles , squares and circles represent workers from North Central, South West and NorthWest Nigeria. Confidence eclipses are drawn at the 95 % level.

568 A.A. Yusuf et al.

HDAA (Figure 2) similar to subspecies of the Asubgroup such as A. mellifera scutellata (Zhenget al., 2010), Apis mellifera intermissa (Crewe andMoritz, 1989) and Apis mellifera saharensis (Hep-burn and Radloff, 1996). Interestingly, the amountsof the queen substance (9-ODA) and its precursor(9-HDA) in workers of A. mellifera adansoniiwere higher than those reported for A. melliferascutellata (Zheng et al., 2010), A. melliferaintermissa and A. mellifera saharensis (Hepburnand Radloff, 1996), but lower than those for layingworkers of A. mellifera capensis (Crewe andVelthuis, 1980) (Table III) and Asian honeybeesApis andreniformis , Apis florea , Apis dorsata(Plettner et al., 1997), Apis cerana and Apisnigrocincta (Keeling et al., 2001). Prior to thisstudy, the ability of workers in African honeybeesto produce substantial amounts of 9-ODA underqueen-right conditions had only been reported inA. mellifera capensis (Zheng et al., 2010) andA. mellifera intermissa (Crewe and Moritz, 1989)workers and had been implicated in the ability ofworkers of these subspecies to rapidly develop intolaying workers (Ruttner and Hesse, 1981). If wecompare the amounts of 9-ODA produced byworkers of A. mellifera adansonii with those ofboth A. mellifera capensis and A. melliferaintermissa , we infer that A. mellifera adansoniiworkers are likely to develop into laying workersmore slowly than those ofA. mellifera capensis butfaster than those of A. mellifera intermissa and

A. mellifera scutellata respectively. However, thisneeds to be verified experimentally in the case ofA. mellifera adansonii .

The presence of substantial amounts of 9-HDAin the head extracts of A. mellifera adansoniiworkers is also of interest since 9-HDA is a pre-cursor of 9-ODA, and high proportions of the twocomponents are referred to as queen-like (Moritzet al., 2004). 9-HDA together with 9-ODA playsan active role in eliciting retinue responses(Slessor et al., 1988), maintaining swarm clusters(Winston et al., 1982), inhibiting queen rearing(Butler and Callow, 1968) and attracting drones(Brockmann et al., 2006). Workers producingqueen-like signals in the presence of a queencould pose a threat to social organisation in colo-nies of other subspecies as seen in A. melliferacapensis workers when they invade colonies ofother subspecies, but function effectively in theirown colonies (reviewed in the study of Neumannand Hepburn 2002). At this point, we do not knowif workers of A. mellifera adansonii will exhibitsimilar behavioural traits.

Queens of A. mellifera adansonii are similar inboth the composition and absolute amounts ofmandibular gland components as those reportedfor other African subspecies like A. melliferacapensis , A. mel l i fera intermissa andA. mellifera scutellata (Crewe, 1987), with 9-ODA and 9-HDA as the main components, slight-ly different from the results on Africanized

Table III. Percentage composition of the major components from head extracts of Apis mellifera adansonii ,A. mellifera capensis , A. mellifera intermissa , A. mellifera saharensis and A. mellifera scutellata workers. Valuesare means±SD from the mean.

Subspecies Percent composition of major components

9-ODA 9-HDA 10-HDAA 10-HDA Reference

A. mellifera adansonii 2.6±4.9 14.7±11.3 24.0±16.9 43.5±23.9 This study

A. mellifera capensis(laying workers)

76.2 7.7 0.5 5.5 Crewe & Velthuis (1980)

A. mellifera intermissa 1.2±1.4 4.2±3.4 0.1±0.3 60.2±11.3 Hepburn & Radloff (1996)

A. mellifera saharensis 1.7±1.1 2.6±0.8 0.4±1.1 44.0±10.0 Hepburn & Radloff (1996)

A. mellifera scutellata Trace 4.7±2.7 25.0±5.0 70.0±7.0 Zheng et al. (2010)

HOB=p -hydroxybenzoate, 9-ODA=9-oxo-2 (E )-decenoic acid, HVA=4-hydroxy-3-methoxyphenylethanol, 9-HDA=9-hydroxy-2(E )-decenoic acid, 10-HDAA=10-hydroxy-decanoic acid and 10-HDA=10-hydroxy-2 (E )-decenoic acid. Values are presentedmean±SD to ease comparisons with previous studies

Mandibular gland pheromones of Apis mellifera adansonii 569

honeybees which on average only had around100 μg of 9-ODA (Pankiw et al. 1996). This isvery important and shows how conservedpheromones of the mandibular glands are inqueens compared to those of workers. Recently,Van Oystaeyen et al. (2014) proposed that a con-served class of queen pheromones (mainly long-chained hydrocarbons and esters in ants, waspsand bumble bees) were responsible for stoppingworker reproduction. The expression of thechemical components of the queen pheromonein the different castes needs to be explored.Since if workers produce some 9-ODA and 9-HDA under queen-right conditions, this maybe a way to reduce the inhibitory effect of thequeen pheromones on themselves and increasethe effect on other workers (Moritz and Crewe,2005), enabling these individuals to get a headstart in the reproductive race if the colonybecomes queenless.

Morphological characters revealed threemorphoclusters that put workers from the Northmuch closer to each other than those from the SouthWest. The morphometric data also provided sup-port for the pheromone data showing workers fromthe North West closer to those from the South.

We found differences in both proportions andabsolute amounts from the head extracts of themand ibu la r g land pheromones and inmorphometric characters of workers from thethree regions sampled, with workers from SouthWest being distinct from North Central and NorthWest respectively. There are two possiblepropositions for this distinction based onpheromone prof i les and morphometr iccharacters. One is that there are possibilities thatthe workers from the South West are A. melliferaadansonii while those from the North Centralcould be hybrids between A. mellifera adansoniiand A. mellifera jemenitica . Both our pheromoneand morphometric data supported earlier work byHepburn and Radloff (1998) which used morpho-metric and multivariate analysis to group honey-bee workers from West Africa into threemorphoclusters based on ecological and climato-logical characteristics. Thus, A. melliferajemenitica in the sahelian and dry tropical zones,A. mellifera adansonii occurred in wet tropicaland equatorial zones and hybrids between

A. mellifera adansonii and A. melliferajemenitica were found in the savannah. However,this does not explain the differences we foundbetween the pheromones of workers from theNorth Central and North West. Since the vegeta-tion in Northern Central is guinea savannah whilstthat from the North West is Sudan savannah(Rabiu et al., 2011).

Our second proposition is that all the workersare A. mellifera adansonii , and the differences weobserved in the pheromone profiles andmorphometrics are as a result of altitudinal,climatological as well as ecological differencesbetween the regions. Indeed, Ruttner (1992) relat-ed the morphological differences among Africansubspecies to altitudinal, climatological as well asecological parameters. Earlier work (seeMutsaers, 1991; Mbah and Amao, 2009;Ebenezer and Olugbenga, 2010) had also revealeddifferent bee forage plants each for the SouthWest, North West and North Central, Nigeria.

We have reported here a detailed account ofthe mandibular gland profiles from head ex-tracts of A. mellifera adansonii workers fromthree ecological regions in Nigeria and haveshown that these are similar qualitatively tothose of A group subspecies. We furtherrevealed compositional and quantitativedifferences in pheromone profiles as well asmorphometric characters among honeybeeworkers from the different regions. Wehypothesise, based on the pheromonal data,that pheromonal development and ovarianactivation would be faster in the Northernpopulations compared to the southern ones.Also, that the values of both parameters willlie between that of A. mellifera capensis andA. mellifera scutellata , two neighbouringsubspecies to the East and South. Our resultsalso provide more insight into the nature and,to some extent, the variability present in theproducts from the exocrine glands ofhoneybees, thus supporting earlier evidencein the study of Crewe (1982) on how compo-sitional variability serves as a key to socialsignals which cannot be ignored due to theirimportance in apiculture especially when thesustainable use of honeybee populations is akey requirement for food security.

570 A.A. Yusuf et al.

ACKNOWLEDGMENTS

The authors acknowledge Alhaji Idris Barau andAdamu Usman of the Beekeeping Extension Society,Zaria, Nigeria, for facilitating the sampling in Nigeria.Femi Kesington of Iwo Apiaries, Babatunde Oreyemiof Oyinladun Beekeepers Society Ijebu-Ode, Manage-ment of Roguwa Farms and other beekeepers forallowing us access to their apiaries. Habibu Aliyu,Ejuailo Dominic and Yusuf Shuaibu of the Departmentof Biological Sciences Ahmadu Bello University Zariaassisted with the morphometric measurements and lo-gistics. Funding for this research was provided in partby the South African National Research Foundation(NRF) and University of Pretoria (2011 and 2012) andfunding to AAY from the DST/NRF SARChI Chair inMathematical Models and Methods in Bioengineeringand Biosciences (M3B2) University of Pretoria (2013).

Contenus de la phéromone de la glande mandibulairechez les ouvrières et les reines d ’Apis melliferaadansonii

abeilles africaines / présence de reines / sécrétions /groupes de phéromones / morphogroupe

Der Gehalt an Mandibeldrüsenpheromon beiArbeiterinnen und Königinnen von Apis melliferaadansonii

Afrikanische Honigbienen / weiselrichtig /Mandibeldrüsensekretion / Pheromoncluster /Morphocluster

REFERENCES

Basile, R., Pirk, C.W.W., Tautz, J. (2008) Trophallacticactivities in the honeybee brood nest—Heaters getsupplied with high performance fuel. Zoology111 (6), 433–441

Brockmann, A., Dietz, D., Spaethe, J., Tautz, J. (2006)Beyond 9-ODA: sex pheromone communication inthe European honey bee Apis mellifera L. J. Chem.Ecol. 32 (3), 657–667

Butler C., R. Callow, (1968) Pheromones of the honeybee(Apis mellifera L.): the “inhibitory scent” of the queen.Proc. R. Entomol. Soc. A: Gen. Entomol., 62–65.

Butler, C.G. (1959) The source of the substance pro-duced by a queen honeybee (Apis mellifera)which inhibits development of the ovaries of theworkers of her colony. Proc. R. Soc. B: Biol. Sci.34 , 137–138

Crewe R.M., (1982) Compositional variability: Thekey to the social signals produced by honeybeemandibular glands. The biology of social insects,318–322

Crewe R.M., (1987) Lability of the mandibular gland signalof three races of African honey bees. Chemistry andBiology of Social Insects, Verlag J. Peperny, Munich,pp.433-434.

Crewe, R.M., Moritz, R.F.A. (1989) Variation in the com-ponents of head extracts of workers and queens of Apismellifera intermissa Buttel-Reepen. Z. Nat.forsch 44 ,590–596

Crewe, R.M., Velthuis, H.H.W. (1980) False queens:A consequence of mandibular gland signals inworker honeybees. Naturwissenschaften 67 ,467–469

Dietemann, V., Pflugfelder, J., Härtel, S., Neumann, P.,Crewe, R.M. (2006) Social parasitism by honeybeeworkers (Apis mellifera capensis Esch.): evidence forpheromonal resistance to host queens’ signals. Behav.Ecol Sociobiol 60 (6), 785–793

Ebenezer, I.O., Olugbenga, M.T. (2010) Pollen characteri-sation of honey samples from north central Nigeria. J.Biol. Sci. 10 (1), 43–47

Hepburn, H.R., Radloff, S.E. (1996) Morphometric andpheromonal analyses of Apis mellifera L. along atransect from the Sahara to the Pyrenees. Apidologie27 , 35–45

Hepburn, H.R., Radloff, S.E. (1998) Honeybees of Africa.Springer Verlag, Berlin, Germany

Johnson R.A., D.W. Wichern, (1998) Applied MultivariateStatistical Analysis, Prentice Hall, Upper Saddle River,New Jersey

Keeling, C.I., Otis, G.W., Hadisoesilo, S., Slessor, K.N.(2001) Mandibular gland component analysis in thehead extracts of Apis cerana and Apis nigrocincta .Apidologie 32 (3), 243–252

Keeling, C.I., Slessor, K.N., Higo, H.A., Winston, M.L.(2003) New components of the honey bee (Apismellifera L.) queen retinue pheromone. Proc. Nat.Acad. Sci USA 100 (8), 4486–4491

Le Conte, Y., Arnold, G., Trouiller, J., Masson, C. (1990)Identification of a Brood Pheromone in Honeybees.Naturwissenschaften 77 , 334–336

Mbah C., A. Amao, (2009) Natural foods and feedinghabits of the African honey bee Apis melliferaadansonii Latrielle (1804) in Zaria, northern Nigeria.Sci. World J. 4 (1)

Moritz, R., Lattorff, H., Crewe, R. (2004) Honeybeeworkers (Apis mellifera capensis ) compete for produc-ing queen-like pheromone signals. Proc. R. Soc. B:Biol. Sci. 271 , S98–S100

Moritz R.F.A., R.M. Crewe, (2005) The interplay ofintracolonial genotypic variation and self-organisation of dominance hierarchies in honey-bees, in: Hemelrijk C. (Ed.), Self-organisationand Evolution of Social Systems, Cambridge Uni-versity Press, pp. 36–49.

Mandibular gland pheromones of Apis mellifera adansonii 571

Mutsaers, M. (1991) Bees in their Natural environment inSouth Western Nigeria. The Nigerian Field 56 , 3–18

Neumann, P., Hepburn, H.R. (2002) Behavioural basis forsocial parasitism of Cape honeybees (Apis melliferacapensi s). Apidologie 33 , 165–192

Pankiw, T., Winston, M.L., Plettner, E., Slessor, K.N.,Pettis, J.S., Taylor, O.R. (1996) Mandibular glandcomponents of European and Africanized honey beequeens (Apis mellifera L. ). J. Chem. Ecol., 22 (4): 605–615

Pirk, C.W.W., Sole, C.L., Crewe, R.M. (2011) Pheromones.In: Radloff, S.E. (ed.) Hepburn R, pp. 207–214. Hon-eybees of Asia, Springer, Heidelberg, Dordrecht

Plettner, E., Otis, G.W., Wimalaratne, P.D.C., Winston, M.L.,Slessor, K.N., Pankiw, T., Punchihewa, P.W.K. (1997)Species- and Caste-Determined Mandibular Gland Sig-nals inHoneybees (Apis ). J. Chem. Ecol. 23 (2), 363–377

Rabiu, M.M., Gudlavalleti, M.V.S., Gilbert, C.E.,Sivasubramaniam, S., Kyari, F., Abubakar, T. (2011)Ecological determinants of blindness in Nigeria: theNigeria national blindness and visual impairment sur-vey. S. Afr. Med. J. 101 , 53–58

Ruttner, F. (1988) Morphometric Analysis and Classifica-tion. In: Ruttner, F. (ed.) Biogeography and Taxonomyof Honeybees. Springer Verlag, Berlin

Ruttner, F. (1992) Naturgeschichte der Honigbienen.Ehrenwirth, München

Ruttner, F., Hesse, B. (1981) RassenspezifischeUnterschiede in Ovarentwicklung und Eiablage vonweisellosen Arbeiterinnen der Honigbiene Apismellifera L. Apidologie 12 (2), 159–183

Simon, U.E., Moritz, R.F.A., Crewe, R.M. (2001) Theontogenetic pattern of mandibular gland componentsin queenless worker bees (Apis mellifera capensisEsch.). J Insect Physiol 47 (7), 735–738

Slessor, K.N., Kaminski, L.-A., King, G.G.S., Borden, J.H.,Winston, M.L. (1988) Semiochemical basis of the reti-nue response to queen honey bees. Nature 332 , 354–356

Slessor, K.N., Winston, M.L., Le Conte, Y. (2005) Phero-mone communication in the honeybee (Apis melliferaL.). J. Chem Ecol 31 (11), 2731–2745

Spannhoff , A. , Kim, Y.K. , Raynal , N.J .M. ,Gharibyan, V., Su, M.B., Zhou, Y.Y., Li, J.,Castellano, S., Sbardella, G., Issa, J.P.J. (2011)Histone deacetylase inhibitor activity in royaljelly might facilitate caste switching in bees.EMBO reports 12 (3), 238–243

Tan, K., Yang, M., Wang, Z., Radloff, S.E., Pirk, C.W.(2012) The pheromones of laying workers in twohoneybee sister species: Apis cerana and Apismellifera . J. Comp. Physiol. A 198 (4), 319–323

Van Oystaeyen, A., Oliveira, R.C., Holman, L., vanZweden, J.S., Romero, C., Oi, C.A., d’Ettorre, P.,Khalesi, M., Billen, J., Wäckers, F. (2014) ConservedClass of Queen Pheromones Stops Social InsectWorkers from Reproducing. Science 343 (6168),287–290

Winston, M., Slessor, K., Willis, L., Naumann, K.,Higo, H., Wyborn, M., Kaminski, L. (1989) Theinfluence of queen mandibular pheromones onworker attraction to swarm clusters and inhibitionof queen rearing in the honey bee (Apis melliferaL.). Insectes Soc 36 (1), 15–27

Winston, M.L., Slessor, K.N., Smirle, M.J., Kandil, A.A.(1982) The influence of a queen-produced substance,9HDA, on swarm clustering behaviour in the honeybeeApis mellifera L . J. Chem. Ecol. 8 , 5

Zheng, H.-Q., Dietemann, V., Crewe, R.M., Hepburn, R., Hu,F.-L., Yang, M., Pirk, C.W.W. (2010) Pheromonal pre-disposition to social parasitism in the honeybee Apismellifera capensis . Behav. Ecol. 21 (6), 1221–1226

572 A.A. Yusuf et al.